The proposed work might consist of five various sections. They are (1) the Perception layer, (2) Attacks Characteristics, (3) Energy efficient multi-hop clustering, (4) Novel clustering approach (PSO based clustering), and (5) Adaptive link selection with improved AODV protocol. The elaborated explanation of the various sections is discussed below.
In Fig. 5, the proposed model is described with functional blocks. The major section of the proposed models is Attack construction and characteristics, Clustering model (Energy Efficient Multi-hop Clustering), Novel Clustering Model (PSO based clustering) and power model with Improved AODV algorithm. All the sections are briefly explained below. In the performance analysis section the major parameters which are concentrated in the simulation are Overhead calculation, Message delivery ratio calculation, Packet loss calculation, Lifetime calculation and spectrum utilization rate calculation.
A. Perception layer
Perception layer is otherwise known as the primary layer to collect the data. It has two sub-sections, namely, perception network & perception node. Perception network guides the control data, whereas perception node controls the data. This layer is affected by the following attacks, namely, eavesdropping, hub catch, malicious attacks, fake node attack & timing attack. These attacks adversely affect the performance regarding authentication, integrity and confidentiality. The network layer provides universal access to the previous layer & regulates information security and network transmission. Denial of Service (DoS) attack, exploit attack, storage attack & Man in Middle are various attacks affecting the network layer's performance authentication, confidentiality, availability and integrity.
B. Attacks Characteristics
This section clarifies some of the attacks that critically affect the network’s performance.
Denial of Service (DoS)
This attack targets the performance in terms of availability, network protocol, computational resources. The signature of these attacks is exhausting memory, bandwidth & time of computing. This attack is the prime reason for generating a software bug and configuration defect. Below Fig. 6 depicts a DoS attack [25]. Hence, IoT frameworks are risk to above mentioned attacks, because of the accompanying reasons [26]. IoT networks are at risk and DoS attacks because of the following reasons.
• Hackers can make a physical attack easily because IoT segments are unattended on the grounds.
• IoT utilizes wireless communication, thus leading to various attacks.
• Interfacing any gadget at any point of time with the network is feasible, because of which an unauthorized entry is also possible in the network.
• IoT environment has limited resources such as bandwidth and energy.
• XExecuting complex security arrangements in parts which has limited resources may restrain the proficient functions of the gadget.
Figure 6 depicts IoT innovations would continue to develop and strengthen security measures to build and maintain clients’ trust. Privacy guarantees concealing private information &be accountable for whatever occurs with that information. Integrity is the capability by virtue of which no intruders can change the unique data from any particular message during transmission process.
Authorization is a controlled permission to network users to use network resources and at the same time safeguards the availability among other users. Authentication guarantees legitimacy of the message by having a source character to be confirmed to the goal. Confidentiality is the ability to check whether the message in transfer mode or rest mode can be either performed by source or destination.
Privacy-preserving Lightweight cryptography
Privacy-preserving Lightweight cryptography is required for IoV framework because of following reasons.
• Efficient end-to-end communication: For accomplishing end-to-end security, the end hubs must essentially utilise a symmetric key algorithm for their execution. In the case of an obliged IoV, PPLWC operation must be with reduced source utilization. PPLWC requires less energy.
• Relevance to resource-constrained devices (RCD): Compared to conventional primitives, PPLWC primitives occupy less space. Hence it has the ability to provide extra gadget connectivity with RCD. PPLWC is one of the primary segments of cryptography utilised for enlarging quick and efficient cryptographic procedure for resource-based territories. This can replace conventional computations while accomplishing a sufficient degree of security.
The salient PPLWC characteristics are mentioned below:
• During the execution procedure, it requires light key size.
• Has a relatively low memory prerequisite.
• It requires little execution time.
• In case of a heavyweight solution, this requires lesser amounts of resources.
The main goal of all Privacy-preserving Lightweight Cryptography calculations is to create a central ground among the exhibition, results and cryptography capability of the calculation. Accordingly, this kind of cryptographic calculation is primarily coordinated to give security-based cryptographic answers for the IoV assets during the on-going condition. PPLWC requires small block size (32, 48, 64 bits) in contrast to the conventional techniques (64, 124 bits). Under all cases, PPLWC requires a block size of less than 96 bits. The most reduced level block size determined from NIST is 112 bits. In ISO/IEC 29192; the target stage-based Privacy-preserving lightweight properties are detailed below.
• The Privacy-preserving lightweight properties are assessed by the accompanying measures for equipment executions. They are vitality utilization and the chip size of the gadget utilized.
• Privacy-preserving Lightweight properties-based programming usage contains a smaller number of codes and RAMS size of the gadget.
• Privacy-preserving Lightweight properties-based programming usage contains a smaller number of codes and RAMS size of the gadget.
C. Energy efficient multi-hop clustering
In this section a novel model for energy efficient multihop clustering method is discussed for showing some sort of improvement in the performance of the VANET. The main aim of the framework proposed is used for connecting each and every vehicle to the gateway-RSU (Road Side Unit) via the internet. In this way, multi-hop neighbors can share data & form clusters. The parameters chosen for the proposed framework are node connectivity, the distance among nodes, link stability and relative speed. Based on low mobility rate, the master cluster is primarily chosen. The cluster stability is enhanced or maintained by properly selecting the slave cluster head.
A significant layer is hub enrolment, neighbour choice, MCH & SCH selection and maintenance. The outcome parameters are group cluster members & overhead, message delivery ratio & latency. Having discussed the parameters for master head selection in the previous section, the parameters for slave head cluster are vehicle cost, vehicle residual longevity and vehicle lifetime value, which assist in making a productive cluster network.
Algorithm for multi-hop clustering:
Input – Initially G represent a group of 30 nodes.
Output – The quantity of master heads be MH
N be the Number of nodes is represented as n greater than Zero (i.e., n > 0)
The network's tolerance value for the supplied matrix be M
Repeat this process for k = 1 to G.
Step 1 – Random matrix division occurs at the beginning
stage such that RM0 ϵ Mfc
Step 2 – Determine the threshold and distance ratios.
Step 3 – If each node's energy is same, then
Step 4 – If matrix (mi) > Tho,
Decide mi as MH
New matrix = matrix (MH)
else step 5
Step 5 – If matrix (mi) < new matrix then
Decide backup MH
Step 6 – Call the PSO algorithm to discover the shortest path
Step 7 – Update the forwarding table with the addition of the
security and energy updates.
Step 8 – To locate each MH's closest hop.
Step 9 – Calculating the Node Distance be ND and Trust
Score be TS
Step 10 – The network's nodes' combined node distance and
trust scores are calculated.
Step 11 – Choose the Node with the highest Trust Value and
the shortest overall distance.
D. Novel Clustering Approach (Particle Swarm Optimization (PSO))
The direction of the work flow is mentioned in the novel clustering approach is shown in Fig. 8. Particle Swarm Optimization (PSO) is a method of improvement wherein typical social and natural behavior’s of species have been taken into account in order to compute. PSO is a swarm knowledge process that relies on a population of people that carry out optimization measures designed to improve the betterment of the work. This approach uses a swarm that searches each particle and logs each particle's estimated wellbeing. The particles are then joined according to their coordination speed at that time. It aids the particle in transitioning to a genuine area by taking into account the cost of the increased wellness capacity. The best location increases the optimized performance for recognizing the group head position to reduce overall energy consumption from all adjacent expertise components. Compared to other mathematical and heuristic methodologies, PSO calculation is proficient and has a higher throughput.
Algorithm for PSO optimization:
Step 1 – Initialization
1.1 – For every particle n, P is the swarm population size.
$$for each P, i=1, \dots {N}_{p}$$
$${P}_{i}\left(0\right) \sim Ud({L}_{Bound}, {U}_{Bound})$$
$$f=\left({X}_{n}\right)$$
$${p}_{best}\left(i,0\right)={P}_{i}\left(0\right)$$
$${g}_{best}\left(0\right)=f\left[{P}_{i}\right(0\left)\right]$$
Step 2 – Repeat till best fitness valueis derived.
2.1 – For every particle n,
$$for each P, i=1, \dots {N}_{p}$$
$$rando{m}_{num1}.rando{m}_{num2} \sim U\left(\text{0,1}\right)$$
$$Velocit{y}_{i}\left(Q+1\right)=\omega Velocit{y}_{i}\left(Q\right)+a{c}_{1}. Rando{m}_{num1}\left({P}_{best\left(i,Q\right)}-{P}_{i}\left(Q\right)\right)+a{c}_{2}. Rando{m}_{num2 }(gBest \left(Q\right)-{P}_{i}\left(Q\right))$$
$${P}_{i}\left(Q+1\right)={P}_{i}\left(Q\right)+Velocit{y}_{i}(Q+1)$$
$$if\left(f\left[{P}_{i}\left(Q\right)\right]<f\left[{p}_{best}\left(i,Q\right)\right]\right), then$$
$${Update {P}_{i} as p}_{best}\left(i,Q\right)$$
$$if\left(f\left[{P}_{i}\left(Q\right)\right]<f\left[{g}_{best}\left(Q\right)\right]\right), then$$
E. Adaptive link selection with improved AODV protocol:
Selection of adaptive relay nodes incorporates periodic "hello" packet transmission and interface quality estimation. The AODV protocol determines the most reliable path from source destination. There is an improvisation in the error message section to rapidly address the connection failure with an elective connection of excellent quality.
F. Radio and Energy model
The radio model detailed in the figure is mainly used for calculating the consumed energy of nodes employed during the transmission between the source and destination [44]. Energy is the primary factor for the transmitter and receiver circuits and has a very close relationship with the number of bits (n).
\({E}_{\left(TX/RX\right)}={E}_{elec} \times n\) ----------------------- (1)
Where, Eelec – Energy taken per bit (n) to start the transmitter (Tx) and the Receiver (Rx) Number of bits (n) and the transmission distance (dt), are the two parameters considered for calculating the transmitter energy consumed by the amplifier
\({E}_{temp}=\left\{\begin{array}{c}{\epsilon }_{fs}\times n\times {d}_{t}^{2}, if {d}_{t}<{d}_{th}\\ {\epsilon }_{mp}\times n\times {d}_{t}^{4}, if {d}_{t}>{d}_{th}\end{array}\right.\) ------------------ (2)
Where dth is a threshold value used for determining the fading model. The value of dth is denoted as,
\({d}_{th}= \sqrt{\frac{{\epsilon }_{fs}}{{\epsilon }_{mp}}}\) ----------------------------------------- (3)
Where ϵfs and ϵmp are the constant values, used for representing the basic models, such as free space and multipath model.
Hence the total transmission energy is given as,
\({E}_{TX}= \left\{\begin{array}{c}{E}_{elec}{\times n+\epsilon }_{fs}\times n\times {d}_{t}^{2}, if {d}_{t}<{d}_{th}\\ {{E}_{elec}\times n+\epsilon }_{mp}\times n\times {d}_{t}^{4}, if {d}_{t}>{d}_{th}\end{array}\right.\) ------- (4)
The receiver energy is given as,
\({E}_{RX}={E}_{elec}+L\) -------------------------------------------- (5)
Calculation of transmission power model:
The relation between the transmitter power (Ptrans) and the receiver power (Prec) is given:
\({P}_{rec}= \frac{{P}_{trans}*{G}_{trans}*{G}_{rec}*{\lambda }^{2}}{{\left(4\pi \right)}^{2}{D}^{2}L}\) ------------------------------ (6)
Where,
Gtrans= Transmission gain of antenna, Grec = Receiver gain of antenna, D = Transmitter to receiver distance, 𝛌 = Wavelength
Here the two-ray propagation model employs the ground reflection of the path. This model is able to provide high accuracy for wider range. The equation of this model is;
\({P}_{rec}= \frac{{P}_{trans} * {G}_{trans} * {G}_{rec} * {H}_{trans}^{2} * {H}_{rec}^{2}}{{D}^{4}L}\) ---------------- (7)
Where, Htrans = height of antenna for transmitter, Hrec = height of antenna for receiver.